August 18, 2016 Equity Research Intel Corp. INTC: Intel Developer Forum Days 2 and 3 Outperform / This week we attended the Intel Developer Forum (IDF) in San Francisco. In this note we present our notes from the second and third day of IDF, and for Sector: Semiconductors completeness we have also included our notes and commentary from the first day (which we published earlier this week). Future technology themes that ran Overweight through many of the presentations included Artificial Intelligence and Deep Learning, Autonomous Driving, Virtual Reality, 5G Communications. Intel emphasized its goal of providing end-to-end solutions for many of these major Company Note trends. We think that Intel’s strength in processor technology and breadth of processor offerings, and its commitment to building out its communications 2015A 2016E 2017E EPS Curr. Prior Curr. Prior capabilities position Intel well to benefit from the large growth in computing and Q1 (Mar.) $0.41 $0.42 A NC $0.56 NC data that we expect over the next several years. We are reiterating our Q2 (June) 0.55 0.27 A NC 0.57 NC Outperform rating on Intel; Intel remains our top pick. Q3 (Sep.) 0.64 0.60 NC 0.73 NC Q4 (Dec.) 0.74 0.73 NC 0.85 NC In our view Intel’s Xeon processors dominate the data center. With Xeon Phi and FY $2.33 $2.02 NC $2.70 NC Altera FPGAs (field programmable gate arrays) in addition to Xeon processors, CY $2.33 $2.02 $2.70 Intel has a full portfolio of products to address the accelerator/coprocessor needs FY P/EPS 15.0x 17.3x 13.0x of artificial intelligence and deep learning. Intel’s Core and Atom processors and Rev.(MM) $55,355 $58,135 $64,000 dominant presence in PCs and data center make Intel a key player in virtual Source: Company Data, Wells Fargo Securities, LLC estimates, and Reuters NA = Not Available, NC = No Change, NE = No Estimate, NM = Not Meaningful reality. Intel’s broad range of processors and 5G capabilities position Intel well in V = Volatile, = Company is on the Priority Stock List autonomous driving technology, in our view. Intel presented data showing that 10nm technology will offer good manufacturing cost advantages, in line with the 14nm and prior technology transitions. It is not clear that Intel will be able to get back to a 2-year technology cadence on the 10nm node. The 10nm node will not use EUV (extreme ultra-violet lithography) technology. Intel has developed its first 7nm technology as an all-immersion (no Ticker INTC EUV) technology, though Intel might use EUV at some point on its 7nm node if Price (08/18/2016) $34.97 EUV machines achieve the appropriate manufacturing capability. 52-Week Range: $24-36 After what we view to be a period of relative silence over its foundry efforts, we Shares Outstanding: (MM) 4,754.0 were interested in some foundry announcements Intel made on the first day of Market Cap.: (MM) $166,247.0 IDF. Spreadtrum is designing on Intel’s 14nm foundry platform. LG will be using S&P 500: 2,187.02 Intel’s 10nm foundry process for a mobile chip platform. ARM will be working Avg. Daily Vol.: 22,907,200 with Intel to offer ARM IP on Intel’s 10nm process. Intel’s Altera division Dividend/Yield: $1.04/3.0% announced that it will deliver its first 14nm FPGA (field programmable gate LT Debt: (MM) $24,053.0 array) product next quarter. LT Debt/Total Cap.: 28.2% Valuation Range: $40.00 to $50.00 ROE: 8.7% Our valuation range is based on an approximate multiple of 15-18.5x our FY2017 3-5 Yr. Est. Growth Rate: 13.0% EPS estimate of $2.70. This is within the range in which Intel has traded in the CY 2016 Est. P/EPS-to-Growth: 1.3x past. Company-specific risks include competition from AMD and rising capital Last Reporting Date: 07/20/2016 spending. Sector risks include the possibility of decelerating growth in PC shipments. Source: Company Data, Wells Fargo Securities, LLC estimates, and Reuters David Wong, CFA, PhD, Senior Analyst Investment Thesis: (212) 214-5007 Intel has a dominant position in a market we view as attractive (microprocessors [email protected] for PCs). We think Intel has a strong lineup of products that will help it gain Amit Chanda, Associate Analyst microprocessor market share. Intel also has a history of strong roadmap execution, (314) 875-2045 enabling it to maintain a leadership position in microprocessor technology. [email protected] Keith Kan, CPA, Associate Analyst (212) 214-5066 [email protected] Joy Zhang, Associate Analyst (212) 214-8017 [email protected] Please see page 12 for rating definitions, important disclosures and required analyst certifications All estimates/forecasts are as of 08/18/16 unless otherwise stated. Wells Fargo Securities, LLC does and seeks to do business with companies covered in its research reports. As a result, investors should be aware that the firm may have a conflict of interest that could affect the objectivity of the report and investors should consider this report as only a single factor in making their investment decision. WELLS FARGO SECURITIES, LLC Semiconductors EQUITY RESEARCH DEPARTMENT This week we attended the Intel Developer Forum (IDF) in San Francisco. Future technology themes that ran through many of the presentations included: Artificial Intelligence and Deep Learning Autonomous Driving Virtual reality 5G Communications. Intel emphasized its goal of providing end-to-end solutions for many of these major trends. We think that Intel’s strength in processor technology and breadth of processor offerings, and its commitment to building out its communications capabilities position Intel well to benefit from the large growth in computing and data that we expect over the next several years. We are reiterating our Outperform rating on Intel; Intel remains our top pick. Intel’s Xeon processors dominate the data center. With Xeon Phi and Altera FPGAs (field programmable gate arrays) in addition to Xeon processors, Intel has a full portfolio of products to address the accelerator/coprocessor needs of artificial intelligence and deep learning. Intel’s Core and Atom processors and dominant presence in PCs and data center make Intel a key player in virtual reality, in our view. Intel’s broad range of processors and 5G capabilities position Intel well in autonomous driving technology, in our opinion. The following are our notes from a number of sessions that took place in the second and third days of IDF. For completeness we are also including our notes from the first day which we published earlier this week. 2nd Day Keynote - Murthy Renduchintala, Intel President 3 themes Pervasive computing will drive a proliferation of data Compute, analytics and storage will be distributed throughout the network. o For example, autonomous vehicles can collect large amounts of information that can be shared with other vehicles close by. The smart future enabled by connectivity o With the shift to 5G compute and communications will converge o Intel plans to play a role in end-to-end deployment of 5G . 5G will enable the interconnectedness of everything . 5G will not be a single homogenous network. The key to 5G is that all the different networks are managed in a seamless manner. Data Center Machine learning o Machine learning is simple math applied at massive scale . This is a small part of overall data center use today but one of the fastest growing forms of workload . Machine learning is a subset of machine learning in which there is more than one level of computation between the input and output. Today deep learning typically involves 2, 4, or 8 levels of computation. Intel estimates that in 2015, About 7% of all servers were used for machine learning and deep learning. Within this 7% Intel believes that: o 95% used Intel processors exclusively (presumably, in a small number of cases, including Xeon Phi coprocessors) o 2.5% used Intel processors together with GPU-based coprocessors (primarily, we believe, Nvidia’s Tesla products) o 2.5% used other types of processors, such as SPARC or Power. o We think these numbers are at least qualitatively correct in indicating that Intel dominates the server processor space and that a relatively small number of servers use coprocessors at all. Intel believes that Xeon Phi has an advantage in that it is inherently scalable – that many Xeon Phi chips can be linked together to work on a given task. 2 WELLS FARGO SECURITIES, LLC Intel Corp. EQUITY RESEARCH DEPARTMENT . Machine learning and deep learning involve training and scoring. In training large amounts of data are analyzed to create a model. o Training is compute-intensive. Intel’s Xeon Phi offers good performance and scalability (more than one Xeon Phi can work together) for training. In scoring trained models are applied to new data to generate predictions – this is the testing and actual use of the learning. o Xeon processors can be used for scoring. o An Altera FPGA can work with a Xeon processor for scoring. o Intel’s acquisition of Nervana gives Intel additional software and expertize in the deep learning space. Accelerators o Computational acceleration can be achieved in several ways: . Within the processor instruction set, by having special instruction for specific tasks. For example in the AVX (advanced vector extensions) instructions accelerate floating point calculations. On the processor die, by including specialized circuity (IP – intellectual property - blocks). For example Intel’s Quick Sync Video circuitry accelerates video. Within the processor package - separate accelerator chips can be packaged together with the processor Intel has begun offering processors packaged together with Altera FPGAs.